Analysis of heart rate variability during exercise stress testing using respiratory information

This paper presents a novel method for the analysis of heart rate variability (HRV) during exercise stress testing enhanced with respiratory information. The instantaneous frequency and power of the low frequency (LF) and high frequency (HF) bands of the HRV are estimated by parametric decomposition of the instantaneous autocorrelation function (ACF) as a sum of damped sinusoids. The instantaneous ACF is first windowed and filtered to reduce the cross terms. The inclusion of respiratory information is proposed at different stages of the analysis, namely, the design of the filter applied to the instantaneous ACF, the parametric decomposition, and the definition of a dynamic HF band. The performance of the method is evaluated on simulated data as well as on a stress testing database. The simulation results show that the inclusion of respiratory information reduces the estimation error of the amplitude of the HF component from 3.5% to 2.4% in mean and related SD from 3.0% to 1.7% when a tuned time smoothing window is used at an SNR of 15 dB. Results from the stress testing database show that information on respiratory frequency produces HF power estimates which closely resemble those from the simulations which exhibited lower SD. The mean SD of these estimates with respect to their mean trends is reduced by 84% (from 0.74×10−30.74×10−3 s−2 to 0.12×10−30.12×10−3 s−2). The analysis of HRV in the stress testing database reveals a significant decrease in the power of both the LF and HF components around peak stress.